This article in Nature calls for “the entire concept of statistical significance to be abandoned.” I’m up for that. It’s almost funny that R. A. Fisher, who did more than anyone to further the cause of the significance test, never wanted us to draw a line with p-values. As far as I understand it, the current mess is the result of textbook authors blending Fisher’s methods with the hypothesis testing approach of Neyman and Pearson (which does produce a choice between two alternatives).
I also like the authors proposed alternative vocabulary of compatibility. I’m going to try it out. There’s a cynical part of me though that says the yes/no appeal of the status quo is going to be hard to shift. Dichotomies have a strong gravitational pull. I looked at this in a recent post for Significance Magazine: An average understanding